# US 2025/0147935 A1 · Method and system for synthesizing data gravity within an incomplete duplicate instance of a very large database

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0147935 A1
- Document ID: US20250147935A1
- Status: Published application
- Kind: A1
- Application number: 18/387,194
- Filed: 2023-11-06
- Published: 2025-05-08
- Earliest claimed: 2023-11-06
- Assignees: JPMorgan Chase Bank, N.A.
- Applicants: JPMorgan Chase Bank, N.A.
- Inventors: Lance ZHANG
- Web page: https://patentgrep.com/patent/US20250147935A1

## CPC

- G06F 16/214: {Database migration support}
- G06F 16/2358: {Change logging, detection, and notification (replication G06F16/27)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A system comprising a processor that may be configured to: monitor an output of a feed that outputs a respective log of each activity that occurs on a source database platform; evaluate the output of the feed against an update session taxonomy to determine whether the respective log indicates a completion of a first update session of first updates to a first source table of a source instance of a live database; when the respective log indicates the completion of the first update session, search a historical database activity repository to detect each respective log that corresponds to the first update session; identify a corresponding data change for each respective log that corresponds to the first update session; and perform a first set of data changes on a copy of the first source table.

## Claims

### Claim 1

A system for synchronizing a copy of at least one first source table of a live database with a source instance of the at least one first source table, the system comprising:
a processor; and
memory storing instructions that, when executed by the processor, cause the processor to perform operations comprising:
monitoring an output of a feed that outputs a respective log of each activity that occurs on a source database platform;
evaluating the output of the feed against an update session taxonomy to determine whether the respective log indicates a completion of a first update session of first updates to the at least one first source table of the source instance of the live database, wherein the first update session comprises a first set of data changes, and wherein the source database platform comprises the source instance;
searching a historical database activity repository to detect each respective log that corresponds to the first update session, when the respective log indicates the completion of the first update session;
identifying a corresponding data change classification for the first update session, based on the update session taxonomy and corresponding logs of the first update session;
determining the first set of data changes based on the corresponding data change classification and the corresponding logs of the first update session; and
applying the first set of data changes on the copy of the at least one first source table, wherein the copy of the at least one first source table is stored within a target database platform.

### Claim 2 (depends on 1)

The system of claim 1, wherein the instructions, when executed by the processor, cause the processor to perform further operations comprising: restoring a connection to the output of the feed,
wherein, when the feed is not offline, each respective log is generated and outputted by the feed immediately after each activity is completed, and
wherein, when the feed is offline, the output of the feed is buffered and the buffered output of the feed is outputted immediately after the restoring.

### Claim 3 (depends on 1)

The system of claim 1, wherein the at least one first source table comprises a set of frequently accessed tables.

### Claim 4 (depends on 1)

The system of claim 1, wherein each respective log that corresponds to the first update session is determined by an artificial intelligence and machine learning (AI/ML) model that has been trained based on each distinct set of update session logs that has been categorized as a distinct type of update session.

### Claim 5 (depends on 4)

The system of claim 4, wherein the update session taxonomy is created and refined by the AI/ML model based on update session training data.

### Claim 6 (depends on 4)

The system of claim 4, wherein when the instructions are executed by the processor, the evaluating further comprises: utilizing the AI/ML model to determine whether the respective log indicates the completion of the first update session.

### Claim 7 (depends on 5)

The system of claim 5, wherein the instructions, when executed by the processor, cause the processor to perform further operations comprising: retraining the AI/ML model based on a set of historical database activity logs that has been categorized as a new type of update session.

### Claim 8 (depends on 1)

The system of claim 1, wherein when the instructions are executed by the processor, the applying further comprises:
transforming the first set of data changes into a first set of corresponding steps; and
incorporating the first set of data changes into the copy of the at least one first source table by having the first set of corresponding steps executed with respect to the target database platform.

### Claim 9 (depends on 1)

The system of claim 1,
wherein the respective log indicates the completion of the first update session, after the update session taxonomy has been utilized to determine that a set of previously generated logs indicate that the first update session exists; and
wherein an alert is issued when the completion of the first update session has not been detected after a predetermined duration of time elapses subsequent to a determination that the set of previously generated logs indicate that the first update session exists.

### Claim 10 (depends on 1)

The system of claim 1, wherein the instructions, when executed by the processor, cause the processor to perform further operations comprising:
determining that the respective log comprises at least one from among a set of database activities that includes insert, update, delete, collect statistics and begin transaction; and
storing, within the historical database activity repository, the respective log after the feed outputs the respective log.

### Claim 11

A method for synchronizing a copy of at least one first source table of a live database with a source instance of the at least one first source table, the method comprising:
monitoring an output of a feed that outputs a respective log of each activity that occurs on a source database platform;
evaluating the output of the feed against an update session taxonomy to determine whether the respective log indicates a completion of a first update session of first updates to the at least one first source table of the source instance of the live database, wherein the first update session comprises a first set of data changes, and wherein the source database platform comprises the source instance;
searching a historical database activity repository to detect each respective log that corresponds to the first update session, when the respective log indicates the completion of the first update session;
identifying a corresponding data change classification for the first update session, based on the update session taxonomy and corresponding logs of the first update session;
determining the first set of data changes based on the corresponding data change classification and the corresponding logs of the first update session; and
applying the first set of data changes on the copy of the at least one first source table, wherein the copy of the at least one first source table is stored within a target database platform.

### Claim 12 (depends on 11)

The method of claim 11, further comprising: restoring a connection to the output of the feed,
wherein, when the feed is not offline, each respective log is generated and outputted by the feed immediately after each activity is completed, and
wherein, when the feed is offline, the output of the feed is buffered and the buffered output of the feed is outputted immediately after the restoring.

### Claim 13 (depends on 11)

The method of claim 11, wherein each respective log that corresponds to the first update session is determined by an artificial intelligence and machine learning (AI/ML) model that has been trained based on each distinct set of update session logs that has been categorized as a distinct type of update session.

### Claim 14 (depends on 13)

The method of claim 13, wherein the update session taxonomy is created and refined by the AI/ML model based on update session training data.

### Claim 15 (depends on 13)

The method of claim 13, wherein the evaluating further comprises: utilizing the AI/ML model to determine whether the respective log indicates the completion of the first update session.

### Claim 16 (depends on 11)

The method of claim 11, wherein the applying further comprises:
transforming the first set of data changes into a first set of corresponding steps; and
incorporating the first set of data changes into the copy of the at least one first source table by having the first set of corresponding steps executed with respect to the target database platform.

### Claim 17 (depends on 11)

The method of claim 11,
wherein the respective log indicates the completion of the first update session, after the update session taxonomy has been utilized to determine that a set of previously generated logs indicate that the first update session exists; and
wherein an alert is issued when the completion of the first update session has not been detected after a predetermined duration of time elapses subsequent to a determination that the set of previously generated logs indicate that the first update session exists.

### Claim 18 (depends on 11)

The method of claim 11, further comprising:
determining that the respective log comprises at least one from among a set of database activities that includes insert, update, delete, collect statistics and begin transaction; and
storing, within the historical database activity repository, the respective log after the feed outputs the respective log.

### Claim 19

A non-transitory computer-readable storage medium for synchronizing a copy of at least one first source table of a live database with a source instance of the at least one first source table, the computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform operations comprising:
monitoring an output of a feed that outputs a respective log of each activity that occurs on a source database platform;
evaluating an output of a feed against an update session taxonomy to determine whether the respective log indicates a completion of a first update session of first updates to the at least one first source table of the source instance of the live database, wherein the first update session comprises a first set of data changes, and wherein the source database platform comprises the source instance;
searching a historical database activity repository to detect each respective log that corresponds to the first update session, when the respective log indicates the completion of the first update session;
identifying a corresponding data change classification for the first update session, based on the update session taxonomy and corresponding logs of the first update session; and
determining the first set of data changes based on the corresponding data change classification and the corresponding logs of the first update session.

### Claim 20 (depends on 19)

The computer-readable storage medium of claim 19,
wherein each respective log that corresponds to the first update session is determined by an artificial intelligence and machine learning (AI/ML) model that has been trained based on update session training data that comprises each distinct set of update session logs that has been categorized as a distinct type of update session, and
wherein when the instructions are executed by the processor, the evaluating further comprises: utilizing the AI/ML model to determine whether the respective log indicates the completion of the first update session.
